Abstract: A survey on analytical model-based fault diagnosis of nonlinear systems is presented. The analytical model based nonlinear fault diagnosis methods are summarized into four categories: nonlinear observer-based approaches, filter-based approaches, differential geometry approaches and adaptive learning approaches. The basic ideas and main research results of each approach in recent 10 years are reviewed in detail, and the advantages and the disadvantages of the methods and the hot research topics are introduced. Finally, some application examples are discussed, and some open issues in fault diagnosis of nonlinear systems and its future research directions are pointed out.